The models of neurons used in network simulations have usually been so oversimplified that studies which have attempted to understand patterns of activity solely as a consequence of the interconnections of the neurons have been a dubious value. We plan to simulate small nerve networks with realistic neuron models in order to see if we can determine the level of detail of neuron morphology and channel distribution at which network output patterns converge. We have long used computer simulations to understand the functioning of single nerve cells and the relative contributions of morphology and ionic channels. We have developed a program,CABLE,(available on PC's and Unix workstations) to simulate neurons with complex branching morphology,multiple channel types,and inhomogeneous channel distribution. A good example of its use in the interpretation of experimental results on failure and recovery of action potential propagation due to partial demyelination and new node formation in multiple sclerosis. We plan to refine our present computer software in order to make the nerve sell simulation task much more accessible to non-specialists in numerical methods. The neuron simulation software will be extended to incorporate several new features,the most important ones being (a),replacement of the current method of spatial discretization by specification of nerve morphology in terms of the component parts of the neuron (e.g. soma, axon, and dendritic tree), (b), addition of synaptic connections between individual neurons, and (c), use of a high level model description language to allow the user to define other physiological mechanisms and insert them at any set of locations on the neuron.